Incorporation of 1 8 0 2 into Thymidine 5 -Aldehyde in Neocarzinostatin Chromophore-damaged DNA*

Strand scission of DNA by the chromophore of neo- carzinostatin converts the 5”hydroxyl of deoxyribose to a 5’-aldehyde. The origin of the aldehydic oxygen has now been elucidated by mass spectrometry. DNA-associated thymidine 5”aldehyde produced by treat- ment of DNA with neocarzinostatin chromophore in zH2180/‘802 or in 2H2’6011802 was reduced, liberated by nuclease treatment, permethylated, and analyzed by gas chromatography-mass spectrometry. The data clearly show that molecular oxygen is the only source of the 5”aldehydic oxygen. The addition of molecular oxygen at C-5‘, possibly via a reactive form of neocarzinostatin chromophore, must be involved; a carbo- nium ion intermediate at C-5‘ is ruled out.

The nonprotein chromophore of the antitumor antibiotic NCS' produces single-stranded scissions in DNA due to the oxidation of the 5"carbon of nucleosides in DNA to the 5'aldehyde (2, 2). Oxidation at C-5' is also involved in the formation of NCS chromophore-DNA adducts covalently linked to DNA sugars (3,4). The mechanism of this reaction, which requires a reducing agent such as thiol and utilizes I mol of O,/mol of chromophore remains obscure (5)(6)(7). It is not known whether oxygen is involved in the generation of an active species which attacks DNA or in the fixation of nascent DNA lesions, or both (7). It is possible that a radical form of the drug generated by thiol addition to the chromophore abstracts a hydrogen from the 5"carbon of deoxyribose in DNA to form a carbon-centered radical which undergoes subsequent reactions leading to 5"aldehyde formation and strand breakage (1,7). Two possible general mechanisms for 5"aldehyde formation can be envisaged, one involving electron removal from the 5'-carbon to form a carbonium ion which after hydration forms the aldehyde and a strand break The abbreviations used are: NCS, neocarzinostatin; AET, S42-aminoethy1)isothiuronium bromide hydrobromide; HPLC, high pressure liquid chromatography; GC-MS, gas chromatography-mass spectrometry. " with a phosphate at the 3'-end (81, the other due to the addition of Oz to the carbon-centered radical to form a peroxyl radical which subsequently degrades to the aldehyde (7). In the former mechanism the carbonyl oxygen comes from H20, while in the latter it comes from molecular 0,. These mechanisms have now been distinguished by mass spectrometric analysis of the oxidized thymidine product formed in the presence of either H2180 or l 8 0 2 .

MATERIALS AND METHODS
Sonicated calf thymus DNA (7.7 mM nucleotide) was prepared as described (9) and dialyzed against 20 mM sodium citrate buffer at pH 4. Sodium horodeuteride (98 atom % 2H) was from Aldrich. 'HzO NCS Chromophore-Clinical NCS (gift of Dr. W. T. Bradner, Bristol Laboratories) was dialyzed against distilled water and lyophilized. The nonprotein chromophore was extracted with 0.1 N acetic acid in methanol (0.2 m1/1.3 mg of NCS) at 0 "C for 2h. The protein precipitate was pelleted by centrifugation, and the colorless supernatant solution (approximately 0.3 mM chromophore) was decanted and stored at -70 "C.
Isolation of Deuterium-labeled Thymidine Produced by Reduction of Thymidine 5"Aldehyde Generated by DNA-NCS Chromophore Reaction-The following components were mixed and lyophilized in the dark (standard reaction): 65 pl of DNA, 150 pl of 10 mM EDTA at pH 4, 85 p1 of 20 mM sodium citrate at pH 4, 500 pl of HzO, and 195 pl of 0.32 mM NCS chromophore solution. To the dried material, 23.1 mg of Tris powder, pH 7.2, (made by premixing 7.02 g of Trizma HCI (Sigma) and 0.67 g of Trizma base (Sigma)) and 3.1 mg of sodium borodeuteride were added, and oxygen was flushed through the septum cap into the vessel for approximately I/i h. Five hundred p1 of 1 mM AET in 'H,I6O was injected into the vessel to initiate the reaction.
The final concentration of each ingredient was as follows: DNA (1 mM), sodium citrate (6 mM), EDTA (3 mM), NCS chromophore (0.125 mM), Tris buffer (300 mM), and sodium borodeuteride (150 mM). The final pH of the mixture was 8.0. To ensure complete reduction, the reaction was maintained at room temperature for 1 h before digestion of the DNA with DNase I (Worthington) and SI endonuclease (New England Nuclear) at 37 "C overnight as described (1). The released thymidine was isolated by reverse phase HPLC using a pE3ondapak C18 column (Waters Associates) (10 pm, 0.39 X 30 cm) as described before (I), except that distilled water was the eluent at a flow rate of 2 ml/min. Thymidine eluted as a single symmetrical peak at approximately 22 min. These fractions were rechromatographed on a Microsorb C18 column (Rainin Instrument Co., Inc.) with a flow rate of 1 ml/min of distilled water and elution of the thymidine peak at 80 min. The lyophilized samples were extracted with methanol, and the thymidine was isolated by evaporation of the methanol under reduced pressure.
Test for Nonspecific 1fiO/180 Exchange-Standard reaction conditions were used except that the sodium borodeuteride (3.1 mg) was not present during the drug scission reaction but was added as solid to the lyophilized reaction mixture after the reaction. zHz180 (0.5 g) was then introduced to initiate the reduction of 5"aldehyde groups and to permit the measurement of the extent of l80 exchange into the carbonyl oxygen prior to the reduction. After lyophilization, the DNA was digested by nucleases in HzO and the thymidine product was isolated by HPLC as described above.
Preparation of Thymidine from NCS Chromophore-DNA Reaction Carried Out in 2H2180-The method is essentially the same as that in 'Hzl'O except that 0.5 g of 'Hz"0 was used. After lyophilization, the DNA was digested in normal water before isolation of the thymidine by HPLC.
Preparation of Thymidine from NCS Chromophore-DNA Reaction Carried Out in 180z--Standard reaction components were mixed in a 5-ml pear-shaped flask and lyophilized in the dark. The residue was

9975
Incorporation of 1 8 0 2 into Thymidine 5"Aldehyde in D N A dissolved in 500 pl of 1 mM AET in 2H,'60. After three cycles of freezing and thawing under high vacuum torr) to expel dissolved I6O2, 23.1 mg of Tris powder, pH 7.2, and 3.1 mg of sodium borodeuteride were transferred into the flask while the solution was frozen at 77 K. The vacuum was reapplied for another hour to the frozen solution, and "02 was introduced. The solution was thawed and shaken to initiate the reaction. After remaining a t room temperature for 1 h, the DNA was digested and analyzed for released thymidine as described above.
Deriuatization and GC-MS-Samples of thymidine (0.25-2 pg) were permethylated as described (10). The resulting derivatives were dissolved in 25 p1 of CHCL,, and aliquots of between 1 and 5 pl were analyzed by gas chromatography on a 30-m DB-1 fused silica capillary column (J & W Scientific) using a Hewlett-Packard model 5840A gas chromatograph equipped with a splitless injector and flame ionization detector. The temperature was held a t 90 "C for 2 min then linearly programmed to 320 "C at 10 "C/min with a carrier flow (He) of 3 ml/ min. The GC-MS computer system consist,s of a Varian model 3700 gas chromatograph (same conditions as above) modified for direct coupling of the capillary column to the ionization source of a Finnigan-MAT 312 double-focusing mass spectrometer operating in electron impact mode with an ionization potential of 70 eV. A Finnigan-Mat SS-200 data system controls the instrument and acquires, processes, and stores the data. For these experiments, the mass spectrometer was scanned from m/z 50 to 500 at a repetition rate of approximately 1.4 s. Short scans over the molecular ion region (m/z 270-300) were done at a scan rate of 2 s/decade. GC-MS of Reduced Synthetic and NCS Chromophore-generated Thymidine 5"Aldehyde"The source of the 5"aldehydic oxygen in damaged DNA was determined with stable isotopes by reacting DNA and NCS chromophore in 2Hz'80/ I6O2 and 2H260/1802 and examining the mass spectra of the corresponding permethyl derivatives for shifts upward in mass of fragment and molecular ions. Hydration of the aldehyde and other possible reactions (see "Control Studies," below) were minimized by reducing the aldehyde, as it formed during the drug-scission reaction, to a primary alcohol which is incapable of further oxygen exchange. The reduced damage product was then removed from the 5"phosphate end of the break by digestion with DNase I and S1 endonucleases. Unfortunately, these enzyme preparations are contaminated with phosphatases which catalyze the release of variable amounts of unoxidized thymidine from DNA. Reduction with NaB2H4 enabled these two sources of thymidine to be distinguished, since only thymidine originating from NCS chromophoregenerated nucleoside 5"aldehyde will incorporate deuterium. Conditions for reduction and subsequent analysis by GC-MS were evaluated using synthetic thymidine 5'-aldehyde. The mass spectrum of the reduced and permethylated material clearly shows by the shift of m/z 145 to 146 that the deuterium is specifically incorporated in the ribose moiety (Fig. lb). The mass spectrum of the reduced and permethylated DNA-NCS chromophore reaction product (2H2160/'602, Fig. IC) is identical to that obtained for the authentic aldehyde, further substantiating the structure of the strand scission product (I). Thymidine released by contaminant phosphatases gives rise to the ion at m/z 284 and the increase in the relative abundance of m/z 145 (Fig. IC). In this experiment (experiment 1, Table I), labeled thymidine accounted for approximately 78% of the total thymidine recovered.

-, and (5'-2H,180)trimethylthymidine obtained by reaction of NCS chromophore with DNA, reduction with NaB'H,, and permethylation
All values are corrected for the natural abundance of I3C.  Table I) results in incorporation of "0 into only 6% of the reducible thymidine. In this case, incorporation of the label is not due to the NCS chromophore-DNA reaction, but occurs through hydration of the 5'-aldehyde prior to reduction (see below). The data in the table were obtained by short, fast magnetic scans over the molecular ion region (mlz 270-300). Measurements of ion intensities obtained in this manner are generally more accurate and precise than those obtained from full scan mass spectra. Full scan spectra were also recorded to confirm that the oxygen is specifically incorporated in the ribose moiety, the observed changes in the relative ratios of m / z 145, 146, and 148 were essentially identical (following correction for isotopic contribution of I3C) to those reported in Table I for m / z 284, 285, and 287 (data not shown).

(5'-'H,'eO) (5',-'H,''O) (5'-ZH
Control Stdies-Hydration of thymidine 5"aldehyde with solvent water would result in exchange of the 5'-aldehyde oxygen (incorporated as the result of specific interaction of the chromophore of NCS with DNA thymidine) for oxygen from water. Therefore, non-NCS chromophore-induced "0 incorporation could result if 2H21'0 is used as the solvent. Conversely, dilution of the NCS chromophore-induced "0 content could occur when 2H2160 is used as the solvent. Reducing agent was added to the NCS chromophore-DNA reaction mixtures to minimize possible exchange by reducing the 5'-aldehyde to the corresponding alcohol as it is formed, thereby fixing the isotopic abundance of l60 and "0 at this position.
The extent of exchange under these conditions was examined by comparing the relative mol percents of (5'-160,2H)trimethylthymidine to (5""0,2H)trimethylthymidine for NCS chromophore-DNA reactions performed in 'H2l60 or 'H2"0 with reducing agent present during the scission reaction (experiments 1 and 3, Table I) versus reaction in and reduction of the lyophilized reaction mixture in 2H2180 in a subsequent step (experiment 2, Table I). The latter experiment indicates that oxygen from solvent water is incorporated, since the relative mole per cent of (5'-",''O)trimethylthymidine has increased from 0 mol % to 10.4 mol % of the total reducible thymidine, as reflected by the ratio of m/z 287 to 285 (Table I, experiments 1

and 2, respectively).
In the actual incorporation studies, the amount of hydrolytic exchange was only half of this value (see experiment 3, Table  I), possibly because the mole ratio of NaB2H4 to aldehyde is kept very large by reduction of the aldehyde substrate essentially as it is formed. In agreement with earlier experiments in which DNA damage was assayed by the generation of acidsoluble material (5), the formation of thymidine 5'-aldehyde was not affected by the presence of metal chelators, such as diethylenetriamine pentaacetic acid, or by scavengers of reduced forms of 02, such as superoxide dismutase or catalase.

DISCUSSION
Cleavage of DNA by NCS chromophore results in strand breaks having a phosphate at the 3'-end and a nucleoside 5'aldehyde at the 5'-end. Previous work has shown that the formation of these breaks involves molecular oxygen (5)(6)(7). This reaction may proceed by a mechanism (e.g. Scheme 1) analogous to those postulated to account for the ability of various electron affinic compounds (including 02) to sensitize DNA to damage by ionizing radiation (13). Although details of the reactions leading to the formation of such a radical centered on C-5' remain to be elucidated, certain facts are known: 1) 1 mol of thiol is consumed during NCS chromophore activation under anaerobic conditions (7); 2) one molecule of thiol adds to the highly unsaturated portion of the NCS chromophore molecule, and this is probably accompanied by a molecular rearrangement to form an activated (possibly free radical) form of the drug (14). Although oxygen is consumed during drug activation even in the absence of DNA (7), it is not known whether oxygen is indispensable for the initial activation. Although there was no evidence for the incorporation of "0 into NCS chromophore treated with thiol in the presence of "02 (14), the possibility that an unstable oxygenated form of NCS chromophore is the activated species DNA + NCS -chromophore + RSH + 0, Incorporation of I8O2 into Thymidine 5"Aldehyde in DNA cannot be excluded. In this case, X in Scheme 1 might be such a species, e.g. thiol-NCS chromophore-00. or thiol-NCS chromophore-0. In any case, some radical form of the NCS chromophore, bound to DNA by an intercalative mechanism (15), may be viewed as abstracting a hydrogen from C-5' of mainly thymidylate residues in DNA (1, 2) to form a carboncentered radical on deoxyribose in DNA, although other mechanisms not involving radicals have not been eliminated.
While various oxidizing species may convert ionizing radiation-induced carbon-centered radicals on the DNA to carbonium ions (16,17) with subsequent aldehyde formation at the C-5' position, the main reaction between O2 and such a radical results in O2 addition to form a peroxyl radical (18-20). Since our data show clearly that the oxygen of the 5'-aldehyde moiety comes from molecular oxygen, an addition mechanism analogous to that shown in Scheme 1 appears to be involved in the NCS chromophore-induced DNA cleavage reaction. Furthermore, our results eliminate a mechanism involving an unstable enol phosphate intermediate (16,21). Such an enol intermediate might be formed either by elimination of HOO from a peroxyl radical at C-5' or by the abstraction of a proton from C-4' via a disproportionation mechanism as described by Kochetkov et al. (22). On conversion of the enol phosphate intermediate to the aldehyde in 'HzO, a deuterium atom would be incorporated into the product at C-4', contrary to our findings. The absence of evidence for "0 incorporation into the 5"aldehyde from 'Hz'*O solvent also precludes this possibility.
If a peroxyl radical adduct were formed in the NCS chromophore-DNA reaction, it presumably would be reduced to the aldehyde by the thiol present in the reaction. In addition, several possibilities for peroxyl radical decay have been proposed for ionizing radiation-induced DNA damage (17,23), one of which may lead to the formation of formate from the 5'-carbon, a minor product of NCS-induced DNA damage (24). It is also attractive to postulate that an NCS chromophore-bound form of O2 participates in the concerted abstraction of a hydrogen and donation of oxygen as part of a ternary complex that is cleaved to form the hemiacetal precursor of the 5'-aldehyde group. Such a mechanism would also explain why only 1 mol of 0, is consumed (per mol of NCS chromophore) (7), even though it would be involved in both NCS activation and DNA damage fixation. It is possible that the thiol, present at high concentration in the NCS chromophore reaction, is involved in the cleavage reaction. Consistent with this possibility are our findings that nucleoside 5"aldehyde formation is increased, while spontaneous base release is decreased, with increasing thiol level: and that a second mole of thiol is consumed following the consumption of 1 mol of O2 (7). On the other hand, the similarity between hydroxyl group formation in the NCS reaction and that produced in cytochrome P-450 systems, as already noted (7), is further supported by the finding that in both cases the hydroxyl group is derived from 02. Finally, the involvement of oxygen in the formation of an aldehyde a t C-5' appears to be unique, for L. S. Kappen  although other electron affinic compounds such as the nitroaromatic radiation sensitizers can substitute for oxygen in DNA strand scission by NCS chromophore, nucleoside 5"aldehyde is not a product (25). Instead, a gap bounded by phosphate at both the 3'and 5'-ends is produced. No evidence implicating a metal-catalyzed reduction of O2 to a diffusible form of DNAdamaging agent has been found in either reaction.